Search Mailing List Archives

Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[protege-owl] Problems with "rdf:type" constructs

Thomas Russ tar at ISI.EDU
Fri Aug 10 09:39:44 PDT 2007

On Aug 10, 2007, at 7:45 AM, Roberto Alves Gueleri wrote:

> But the problem cannot be solved only by assenting that  
> "MetaClassA" is a subclass of "owl:Class". This would be true if I  
> only wanted to reach the order - or level - of "MetaClassA". But  
> what about the possibility for creating, for example, a  
> "MetaMetaClass", such that I will assert that "MetaClassA" is an  
> occurrence of "MetaMetaClass", as folows:
> ...
> <owl:Class rdf:ID="MetaMetaClass"/>
> <owl:Class rdf:about="#MetaClassA">
>   <rdf:type rdf:resource="#MetaMetaClass">
> </owl:Class>
> ...

I don't think that it matters.  Since both meta classes and meta-meta  
classes are subclasses of owl:Class.  The main issue is that Protege  
needs to know whether a particular name refers to a class or else to  
a non-class individual.  Meta classes and meta-meta-classes, etc. all  
have instances which are classes.  That is the key bit of information  
that Protege needs in order to correctly build its internal data  
structures, and which needs to be specified.

> Cases like this can be viewed in the owl ontology of the OpenCyc:
> where we find many transitions of orders by using "rdf:type"...

But don't forget that Cyc is actually defined using a much more  
expressive language than OWL.  IIRC it uses what is essentially a  
variant of 2nd order predicate calculus.

More information about the protege-owl mailing list